Publicly available AI models still have some way to go before they are production ready. They're still too inconsistent stylistically compared to a 3d model with a few different shape keys or a 2D image with variant layers.
Tale of Khostol is a great example of how your proposal can go wrong. That said, a few custom models with a diet of stylistically similar images could be trained for the task.
Yes, but that type of counter argument was already covered in my post - I didn't say it was all in place and ready to go right now with public models. The typical dismissal of AI use in
future for whatever role is oddly always based on the
now, a total short sighted strawman, but tries very hard not to look to the looming scary tomorrow. It really is short sighted, like someone seeing an early personal computer in the 70s and thinking it'll go nowhere and just a fad because it can't do much right now in their mind. Same with many other industrial or machine revolutions through time in whatever niche. The rate of development and progress of AI is astonishing and only getting faster and more amazing results. It's simply a question of time as it's obvious there's no reason for it not to move into the scene. Why wouldn't it?
Same with normal online content creation (YT, news, whatever), streamers, etc. AI will be a huge factor in all forms of entertainment and will grow rapidly with time. There will always be a place for hand crafted art but it will be like buying hand made furniture now.
It will be used in all forms of marketing, propaganda, government services, commercial services, intelligence, etc. The investment in this is the new bonanza. Because in the end there's no reason not to do it. At least as far as the owners and users of it are concerned.